We created an experimental adaptation of the REF scale (Lenzenweger et al., 1997), in a Likert format for discriminate between frequency and intensity of referential thinking (Rodríguez-Testal et al., 2008).

We try to verify if the Likert format of the REF discriminates between controls and patients, and also in patients with different diagnoses. We predict that there will be differences in frequency and intensity between patients and controls.

Participants: 108 subjects, 40 patients from a private center of clinical psychology, 55% women, mean age = 35.70 (SD = 12.42) and 68 controls selected from the normal population, 50% women, mean age = 36.35 (SD = 12.99).

It was applied a cross design for a correlation method of comparison between groups. All the analysis were accepted at p< .05.

No differences in referential thinking between patients and controls with Likert format in frequency (t = 1.496, P = 1.14), although there were differences in intensity (t = 2.30, p =.023). No significant differences in referential thinking between types of diagnoses with the Likert format (X2 = 6.63, p =. 249).

The Likert format of the REF scale adequately discriminates between patients and controls in intensity but not in frequency. This format doesn't discriminate between different diagnoses. The Likert format induces and overestimates the response.